Steve Chipman

@thinking_slow @ahrefs I asked Grok about a scenario related to this possibility x.com/i/grok/share/7… . Of course, it would have to be tested.

@thinking_slow @ahrefs A possible exception is adding authoritative sameAs entries, which helps with brand disambiguation for brand names with non-unique text strings. This can make a brand more prevalent in search and LLMs vs. near-matching brands.

does adding schema markup help your pages get cited in AI search? probably not 👇 we (@ahrefs) analyzed 6M URLs and found schema is more common on heavily cited pages BUT that's probably correlation, not causation: schema markup is more common on pages with good SEO generally so we tested 1,885 pages that actually added JSON-LD schema and compared their citations before and after verdict: across Google AI Overviews, AI Mode, and ChatGPT, we saw *no measurable difference* in citations (or even a tiny negative effect) schema matters for plenty of reasons in SEO and AI search, but adding schema to your pages is probably not some magic fix for improving your AI citations ✌

A new AI-native CRM vendor chose an ambiguous brand name. It's really best to choose something unique for the reasons stated above.

Example of LLM domination.

Brand Name Optimization (BNO? - because we need another SEO TLA). HubSpot's 2006 thought process behind its brand name has a critical new angle in 2026. If you're starting a new company or rebranding, think about LLM domination.

Here's a content pruning workflow for content that can save you a lot of time: → Claude reads your sitemap, flags posts that appear stale (defunct apps, dated references, rebranded vendors) → Its Ahrefs connector pulls traffic + backlink data for the shortlist → Claude reconciles: which can 404, which need a 301, which should be rewritten because they have a high-value backlink you'd hate to lose A sitemap audit alone misses link equity. A backlink audit alone misses topical decay. You need both.

@DavidGQuaid @RohanHosmani @JoyanneHawkins A faction of Google says this #html" target="_blank" rel="nofollow noopener">developers.google.com/style/headings… It's good practice, but not life-or-death practice.

@RohanHosmani @JoyanneHawkins Both. And I'm not saying it - I'm relaying what google is saying.

99% of SEOs and Web Devs refuse point blank, at the pain of death - to read the SEO starter guide. It resolves 90% of SEO "debates" 1) EEAT is not real - Google Jokes about it 2) H headers can be used in any order 3) You do not need ANY words - 0 words is fine; ergo: there is NO problem with THIN CONTENT 4) PageRank is fundamental Its all in one page - go print this, laminate it and staple it to people's heads

@gothburz Let's get together for golf at your Madera club in late 2034. I'm assuming we'll be FSD'ing to the location.

I am the Director of Strategic Planning at the California High-Speed Rail Authority. I have held this position for seventeen years. In that time I have written four business plans, overseen six revisions, and authored eleven methodology updates. The train has not moved. There is a hard hat on the shelf behind my desk. It was given to me at the Fresno groundbreaking ceremony in 2015. It is still in the cellophane. I use it as a bookend for the business plan binders. There are four binders. They are substantial. The hat holds them upright. In 2008, California voters approved Proposition 1A. San Francisco to Los Angeles. Two hours and forty minutes. Fifty-five dollars per ticket. Ninety-five million annual riders by 2030. Total cost: $33.5 billion. Fifty-three percent said yes. The current cost estimate is $231 billion. I am sometimes asked to provide context for that figure. The state housing shortage is 2.5 million units. At the California median home price, $231 billion would produce 577,000 of them. The average public school teacher in California earns $95,000. $231 billion is every one of their salaries for eight years. The state has a documented wildfire suppression staffing gap. $231 billion would fund 6,400 additional fire crews for a century. I include these comparisons for context. They are not relevant to my work. We have built 119 miles of infrastructure. Columns. Viaducts. Grade separations. You can see them from Highway 99 between Madera and Bakersfield. They stand in rows across land that used to grow things. No track runs on them. No train has touched them. Some of them have graffiti now. I have seen the photographs in the quarterly progress reports. I have not visited. There was an almond grower outside Hanford. She is in our files as Parcel 417, Hanford East. The Authority acquired twelve acres of her property through eminent domain in 2016 for right-of-way clearance. The trees were removed. The soil was graded flat. The right-of-way has been clear for nine years. Nothing has been built on Parcel 417. Her file notes that she attended three public comment hearings between 2014 and 2016. I do not know what she said at those hearings. I know what we said. We called the acquisition "a critical milestone in the project's advancement." I wrote those words for the 2016 Annual Report. They were well-received. Her contact information has been flagged in the Phase 2 preliminary assessment, in case additional right-of-way is required. Phase 2 does not yet exist. Her contact information does. The original completion date was 2020. The current target is 2032. The route has been revised from San Francisco-to-Los Angeles to Merced-to-Bakersfield. One hundred seventy-one miles. I refer to this as Phase 1. The French national rail company, SNCF, joined the project as a consulting partner in 2010. They left in 2011. They used the phrase "political dysfunction," which is diplomatic language for a country that built the Eiffel Tower in two years telling you it cannot build your train. SNCF then went to Morocco and built a high-speed rail line from Tangier to Casablanca. Two hundred miles. Operational by 2018. Seven years. We are in year eighteen. I included the SNCF departure in the 2022 business plan as a "comparative international case study." The lesson I drew was that Morocco has simpler permitting requirements. This is accurate. I did not draw other lessons. The $9.95 billion bond that voters approved costs the state $647 million per year for thirty years. Roughly $20 billion in total repayment. The bond is being serviced on schedule. $647 million leaves the state treasury every year and arrives in accounts associated with a train that does not carry passengers. It has done this since 2010. The bond repayment is the most functional transit system we have built. It moves $647 million a year. On time. Every time. In 2019, Governor Newsom said the project "would cost too much and, respectfully, take too long." He then continued funding it. I appreciated the word "respectfully." It acknowledged the problem without producing an obligation to solve it. My team delivered the 2020 business plan revision the following quarter. It was well-received. The Governor also supported legislation to shield certain cost details from public disclosure. That same year, thousands of pages were removed from the Authority's website. I was not involved in that decision. I was involved in the pages. Last Friday, a television host told the Governor on camera that the project now costs $231 billion. The Governor said, "No, it's not. It's not." The $231 billion figure is from the 2026 draft business plan. Page 47. I wrote page 47. The Governor then said we had gotten the project "back on track." I noted the phrasing. A rail project that has not yet laid operational track is not, in a strict sense, on one. I did not raise this. The ridership projection has been revised from 95 million annual riders by 2030 to 36 million by 2060. I updated that figure personally during the 2024 planning cycle. The ticket price has been revised from $55 to $105. Both figures describe a service that does not yet exist. The State Auditor published a report. The title is: "Flawed Decision Making and Poor Contract Management Have Contributed to Billions in Cost Overruns and Delays in the System's Construction." That is twenty words. We have 119 miles of columns that carry nothing. I read the report carefully. It was thorough. We incorporated its findings into the next business plan revision. The board was scheduled to vote on the latest business plan on April 29th. The vote was delayed. Additional review was requested. I support additional review. My pension vests in 2034. The project's current completion target is 2032. If the train is finished on schedule, my position becomes unnecessary two years before my pension matures. The completion date is determined by the business plan. I write the business plan. The plan says the project should continue. It has always said this. I have never written one that recommended otherwise.

@searchmartin If you look at Ahrefs' Web Analytics > Traffic > Search, you'll see vastly different numbers across sites compared to Overview > Organic Traffic.

Semrush and ahrefs traffic "estimation" numbers are no better than guessing, stop taking business decisions based on them.

It's also about relative token efficiency.

An authoritative leapfrog indicator.

@DavidGQuaid @rustybrick @edwardeachday @seosmarty It's been in NotebookLM for a while, but I've never seen it in the SERPs.

Wait - have you guys heard this??? Google just created a 2 person pod cast (one man and one female voice) @rustybrick @edwardeachday @seosmarty

@Benjamincanks @AlexLathery Yes, for convenience and to avoid an additional sub. But I didn't suspect that something so simple could be inaccurate. Thanks for calling it out.

@SteveChipman @AlexLathery Do you use Ahrefs for website analytics? Is it providing accurate data? I switched to Usermaven because Ahrefs was not showing accurate analytics

I used to be pretty all-in on Ahrefs for keyword rankings, traffic estimates, and all the usual SEO reporting. This year, though, I’ve had a few too many moments where I compare Ahrefs to Google Search Console and just sit there wondering how they can be looking at the same website. The gap is not small. You can look at Google Search Console and think a site is absolutely crushing it. Then open Ahrefs and see very little actual progress. Or sometimes the opposite. At a certain point, it gets hard to treat Ahrefs data as anything more than directional. I’ll still use it for keyword research, competitor analysis, and all of that. But reporting results off Ahrefs data alone just doesn’t make much sense to me anymore. The numbers are too far from reality.

Many of these can be done with any platform. Computer did them better. The pre-built connectors facilitated a number of these. And this is just for starters. Finance - Investment due diligence — reviewed project documents and produced a structured report covering land control, zoning, minerals, and environmental gaps - Infrastructure feasibility analysis — evaluated capital requirements and tax credits for an AI data center project with 'behind-the-meter' power - Internal-use calculator for running scenarios on blends of solar, batteries, and fuel cells for an AI data center - Research and report on best-fit prospective data center tenants and investors by parsing LinkedIn profiles Sales & CRM - Pre-meeting company and contact research — well-researched, contextual information about initial sales call participants and their organization in an easily digestible format - Created a lead qualification skill to research form submitters via the ActiveCampaign connector, produce a report, and conclude whether a submitter should be categorized as an MQL - Did the same as above with the Cal .com connector to determine whether to accept or reject a meeting request - Prospect meeting transcription summary — a better-structured informational format for post-call SDR to AE internal handoffs than what native AI notetaker-generated summaries provide - Advanced analysis of sales opportunity velocity — generated a stage duration analysis for pipeline review meetings - AI-native CRM platform research — evaluated current options and presented a detailed breakdown of features, tradeoffs, and recommendations Marketing & SEO/GEO - Content auditing — pulled live Google Search Console data to identify declining pages and received specific, actionable recovery recommendations for refreshing content - Internal link auditing and building — analyzed existing blog content and generated ready-to-insert sentences with natural anchor text and placement instructions - FAQ generation — identified the optimal question set to add to a blog post based on Google Search Console data - Structured data creation — constructed off-page structured data for custom webpages - Web performance tool configuration guide — researched industry-wide best setup practices for a popular WordPress plugin and produced a formatted PDF reference guide for an internal webmaster - Online tools — created calculators and assessment tools to attract website visitors - Vendor directory web page — researched and built a styled, multi-company directory with overviews, funding data, and links to founder LinkedIn profiles and company pages. Social Media - Wrote platform-ready LinkedIn posts with accompanying images that summarize internally and externally created article content - Recurring content workflows — set up automated article curation from specific sources on a schedule

@SteveChipman @CharlieRegnum @Jason @AravSrinivas Can you give examples of use cases with computer? I built some financial reporting dashboards and Customer account dashboards and burned through hundreds of dollars of tokens. Each refresh takes a ton of tokens as well. Doesn’t feel sustainable to me

Super impressive product

I like Gemini, but Perplexity Computer does everything better. It even gets better results when it calls Nano Banana or Veo, because it smartly enhances your prompts. Today, I use a combination of Computer and Gemini to save on Computer credits on Enterprise Pro. If Computer credits were a third to half the price, that's all I'd use.

@Jason @AravSrinivas Q: why use perplexity over claude or gemini?

@AravSrinivas Congratulations. Computer is a tool that seems to continually exceed expectations.

Perplexity started as a small business tool for ourselves. We had 4 people and no revenue with AI at our fingertips. The pivot to Computer is actually a full circle. Founders are using it to grow companies that matter to the economy and their communities.  It’s rewarding to see it now powering small businesses and startups in big ways. Perplexity is still a startup. We just 5X’ed revenue from $100M to $500M with only 34% growth in team size. 2x revenue growth in 2026 with same small team. And we’re just warming up. Everyone here works at a small business, and everything we build is for people who build.

Does your top-level domain matter for SEO? Data from TopSEOFactors.com suggests that the classic TLDs (.com, .net, and .org) still hold weight in Google’s algorithm. (Keep in mind that there may be causal factors such as longevity and associated domain authority.) If you are launching a new brand, the data suggests that sticking to the 'big three' extensions provides a statistical edge in ranking stability, especially following major core and helpful content updates. Newer TLDs (.ai, .io) may not be the best choices if organic search is a growth lever. The percentage of keywords where a .com, .net, or .org domain is a significant ranking factor has climbed from around 40 percent in 2019 to over 72 percent in 2026. This is, as you would expect, classified as a difficult factor to address because you cannot easily change your domain without side effects. The data suggests that while a .com TLD does not guarantee a top spot, the highest-ranking results are still dominated by these legacy extensions. Here are the charts: topseofactors.com/factor.php?id=…

Google lost ~5% of traffic share in the past 10 months (35.11% → 30.53%). Everyone thinks AI search ate it. Well… ▪️ AI search: 0.22% → 0.26% (+0.04pp) ▪️ Social: 7.67% → 8.24% (+0.6pp) ▪️ Paid: 13.99% → 17.15% (+3.2pp) ^ that’s across ~75k websites in @Ahrefs’ panel. (HINT: visit chatgpt-vs-google(DOT)com to see more data) ... AI search gained almost no traffic share. And it makes sense. AI search is zero-click by nature. It answers questions, it doesn't send traffic. The real winner? Paid. Businesses are losing organic clicks from Google and compensating with ad spend. They have no choice. They still need customers on their websites. So Google pushes AI Overviews, organic traffic drops... and businesses respond by giving Google more money for ads. ..or at least that's my read on the situation. What's yours?